Tag Archives: data science

Getting Mobile Advertising Right: You Have One Opportunity

Mobile Advertising is more like classic Direct Response Marketing than Web Advertising: you have one opportunity for the perfect pitch: for each person, based on who he is and what his is doing—right NOW.

Mobile Advertising is NOTHING Like Web Advertising

Facebook’s IPO, and the myriad analyst remarks on its challenges in the mobile space, has brought the discussion about how to make mobile advertising work back into the limelight. Many have argued that mobile advertising, especially mobile advertising in non-Search apps, has much lower likelihood of success because customers are not in the process of “seeking to buy something”. These arguments are based on the assumption that mobile advertising is like web advertising. This assumption is wrong.

Web advertising (in-text or display ads) offers you the opportunity to present many advertisements to a customer at once on a screen. You can leave these ads up for the duration of the customer’s perusal of the screen or rotate new ads in place over time. In addition, if the customer is logged in (or you have really good cookies) you have high certainty of his or her identity.

Mobile advertising is entirely different. The screen real estate only provides the opportunity for one advertisement. Even worse, you only have a small amount of time (less than two seconds) as your advertisement is “getting in the way” of the customer’s attempt to do something on their smartphone. What you do have in your favor is near certainty of the customer’s identity.

The Approach Needed Solve Mobile Advertising Has ALREADY Been Developed

This is not a new challenge. It is the same situation faced for years when cross-selling products and services to customers from the call center. They had: 100% accurate customer identification, lots demographic and account information on the specific customer, and only a few seconds to offer one compelling promotion before ending the call.

The trick to solving this challenge was to figure out the one ideal promotion to present to each customer based on who he is, what he is currently doing, and the current time-of-year, day-of-week, and time-of-day. Just as important is using the feedback on each to calibrate future promotions to the same customer (to avoid turning advertising into a nuisance), making this more of a Recommendation Engine challenge than an Advertising Engine one. The rewards are enormous: bounty payments for accepted promotions are frequently 100x greater than those for clicked-on ads.

The Technology Exists to Scale This to Mobile

A decade ago, we scaled this model from the 10-transaction-per-second world of call centers to the 10,000-transaction-per-second world of the Internet, generating billions of dollars of value per year. Now is the time to scale this to 1,000,000-transaction-per-second world of mobile to capture tens of billions of dollars in value (luckily we can now grab Big Data technologies off-the-shelf to do this, in the past we had to invent new technologies to scale 100- and 1,000-fold). Mobile, with its “Perfect combination” of validated identity, addressable application data, location awareness and real-time notification services offers an amazing opportunity to take this to the next level.

The Results Would Be Incredible

Imagine this mobile Yelp-like example:

Barney has smartphone and is in the Financial District in Manhattan, Monday through Friday each week. When he installs your app, you get his email and mobile phone that lets you (via sources such as Flurry and PRIZM codes) guess he is likely an affluent male in his mid-thirties. Based on this you may want to advertise local bar Happy Hour promotions when he opens your App between 12pm and 6pm ET on Thursdays. Clicking on the promotion provides a bar code, QR code or confirmation number for redemption with the ad buyer. You can adjust future advertising by tracking redemption rates and controlling for mobile location, day-of-week and time-of-day.

Adding social data to mobile makes this scenario even more valuable. Imagine this mobile Facebook-like example:

Barney has entered lots of information about himself in your App: he is single, he is interested in women, he works at Goliath National Bank (GNB), etc. You can now get incredibly targeted. You can offer a promotion that gives Barney more savings if he brings co-workers from GNB with him. You can now track his response against others based on location, day-of-week, time-of-day and a slew of confirmed individual demographic data (gender, employer, age, etc.) to plan and refine future promotions.

Companies like Facebook, Twitter, Groupon, FourSquare, Yelp and many others have assembled a “treasure trove” of data on customers. Today’s technologies make it easier for companies to parse this data for recommendation and promotion than ever before. Apple and Google make it easy to reach over a billion people worldwide through in-App ads, alerts and notifications. The next step is to map traditional cross-sell models into the mobile space (rather than force-fitting web advertising models).

2020 Challenge: Completely re-invent how we process data (or grow our brains 30x!)

matrix-200pxOn Friday, Russell Garland of the WSJ wrote about the “Data Tsunami” that is coming due to increased volumes of data being generated from everything from the Facebook Social Graph, the next Interest Graph and genomics (just to name the most obvious growth driver). “Tsunami” is probably too small of a word (unless you are talking about Jupiter-scale growth). Take a look at these interesting numbers:

  • The average human brain can take in and remember about one byte per second (two gigabytes over an average life time, including sleep)[1]
  • The amount of data storage in the world in 2000 was rough 300,000 terabytes—about 0.03 “brains’ worth” of storage for every person on Earth[1,2]
  • This amount grew at to approximately 1,200,000 terabytes by 2010—about 90 “brains’ worth” of storage for every person on Earth.[2,3] No wonder we feel so over-loaded with data!
  • By 2020, this will get even more outlandish. We will have 36,000,000,000 terabytes of data—about 2,400 “brains’ worth” of storage for every person on Earth.[2,3]

Managing storage of this volume data will be an interesting challenge for companies like EMC, IBM and Oracle (one aided greatly by Moore’s Law). However, being able to understand it will require complete reinvention of how we process, explore and analyze data.

These new technologies will be as advanced when compared to today’s data warehousing and reporting technologies as the spreadsheet was when compared to manual ledgers. They will use non-linear rule engines and artificial intelligence to find trends and determine which data are most important. They will use new data visualization techniques, leveraging everything from 3D to augmented reality (AR) technology to enable human-scale brains to explore results and conduct analyses. This, in turn, will drive new physical interfaces from the desktop to mobile to even wearables.

It should be a very interesting ride!

Notes:
[1] “How Much Information is there in the World?”, Michael Lesk
[2] “World Population”, Wikipedia
[3] EMC-sponsored IDC study, “The Digital Universe Decade – Are You Ready?”